Organic Fault-Tolerant Control Architecture for Robotic Applications

Size: px

Start display at page:

Download "Organic Fault-Tolerant Control Architecture for Robotic Applications"

Alvin Boone
8 years ago
Views:

1 Organic Fault-Tolerant Control Architecture for Robotic Applications Werner Brockmann, Erik Maehle University of Lübeck Institute of Computer Engineering Director: Prof. Dr.-Ing. E. Maehle Karl-Erwin Großpietsch Fraunhofer-Institut für Autonome Intelligente Systeme Schloss Birlinghoven, Sankt Augustin Karlsruhe, 14./

2 Overview Introduction ORCA Architecture Application Example Methods Conclusion

3 Introduction Motivation Intended applications: autonomous mobile robots in human environments -> unstructured, -> complex control dynamically changing systems environment -> no explicit model -> no explicit fault of the environment model fault-tolerance, safety engineering bottleneck

dynamically changing systems environment -> no explicit model -> no

4 Introduction Organic Systems Organic systems adapt dynamically to environment and malfunctions - uncertainties - unknown environment - unforeseen situations - errors - faults Our approach: Inspiration: Aim: controlled self-organization autonomic nervous system and immune system - detect, react, and adapt to malfunctions - avoid critical system states at any time - low cost implementation and engineering

Inspiration: Aim: controlled self-organization autonomic nervous system and immune system - detect,

5 ORCA- Architecture ORCA - Organic Robot Control Architecture Planning Behaviours OCUs decisional level functional level Reflexes Reflexes Gait-Pattern-Gen. Gait-Pattern-Gen. Gait-Generation / -Selection OCU Perception Proprioception Motor-Controllers Motor-Controller Gait-Pattern-Gen. Gait-Pattern-Gen. Sensor Modules Motors Motor (PWM) (PWM) hardware level BCU = Basic Control Unit OCU = Organic Control Unit

6 ORCA- Architecture OCU-BCU-Interaction Software architecture DAISY (Distributed Architecture of Intelligent SYstems) OCU BCU OCU BCU BCU

7 ORCA- Architecture OCU-Architecture Memory Monitor Reas oner - Monitor: anomaly detection - Memory: short term history - Reasoner: hard real-time determination of a counteraction

8 Application Example Climbing Robot DEXTER (DEXTerous Electric Rover) Technical data: - size 36, cm 3 - weight about 3 kg - 4 active degrees of freedom - passive suction cups - IR- and US-sensors on feet Characteristics: - distributed control system - alternative behaviors - incomplete perception

degrees of freedom - passive suction cups - IR- and US-sensors on feet

9 Application Example Climbing Robot DEXTER (DEXTerous Electric Rover)

10 Methods Methodological Approaches - signals reflecting the health of a signal or a BCU e.g. - noisiness, confidence of output results, - load state; error state - adaptive action selection IF movement=blocked THEN activation of commanded behaviour -=5%, activation of non-commanded behaviour +=5% - learn to treat malfunctions -> learning at BCU- and OCU-level

selection IF movement=blocked THEN activation of commanded behaviour -=5%, activation of

11 Methods Supervised Learning - no formal models OCU - fast adaptation - manageability - controlled self-organization (safety) -> hybrid crisp-fuzzy systems Supervised Process (robot, BCU, OCU,...) - high-dimensional problems -> adaptive filters

(safety) -> hybrid crisp-fuzzy systems Supervised Process

12 Methods Hybrid Crisp-Fuzzy Systems Key features: - granular computing (partitioning of the input space) - mixed crisp and fuzzy processing - normalization of rule-base - rule-based specification as well as learning - very fast computation, hard real-time learning - decomposition to fight complexity - individual attributes for each partition -> tag-bit for guided online-learning -> safety warranty

as well as learning - very fast computation, hard real-time learning - decomposition to fight

13 Methods Adaptive Filters Key features: - complexity reduction of the control space - weighted summation out of history array - learning of linear parameters - minimizing the difference between desired and measured system behaviour - at higher system levels, combination with hybrid crisp-fuzzy systems - used for -> health monitoring of local components -> compensation of small local malfunctions

desired and measured system behaviour - at higher system levels, combination with hybrid

14 Conclusion Conclusion Application: fault-tolerant autonomous mobile robots -> flexible, but safe adaptation to malfunctions (unforeseen events, errors, faults, ) Architecture: mimic autonomic nervous system and immune system -> ORCA-architecture + learning Methods: - health -status - controlled self-organisation -> hybrid crisp-fuzzy systems -> adaptive filters

mimic autonomic nervous system and immune system -> ORCA-architecture + learning Methods:

Diagnosis and Fault-Tolerant Control

Mogens Blanke Michel Kinnaert Jan Lunze Marcel Staroswiecki Diagnosis and Fault-Tolerant Control With contributions by Jochen Schroder With 228 Figures Springer 1. Introduction to diagnosis and fault-tolerant